Download files
Abstract
The ionosphere is an atmospheric region, which includes a significant number of free thermal electrons and ions located at altitudes of 60km to beyond 1000km above the Earth. The ionosphere can act as the media for long-distance propagation of electromagnetic waves. On the other hand, it also simultaneously disturbs the transmission of these radio signals. Therefore, not only the features of the ionosphere need to be thoroughly studied, but also it is necessary to monitor and predict the ionosphere in near real time through effective ionospheric-detecting instruments and models.
Nowadays, it is still a challenging issue to build up accurate and easy-to-use ionospheric models because of the ionospheric variability and lack of very effective instrumentation. The drawbacks of many currently used ionospheric models include: 1) they need to simultaneously combine many different data sources e.g., Global Positioning System (GPS) data, radio occultation (RO) data, data from in situ measurement (i.e., ionosondes, the incoherent scatter radar, the relative ionospheric opacity meter) and the data from the airglow radiances, etc. However, because some of these data sources are difficult to obtain at many target locations at the expected time, it will limit the widespread use of these ionospheric models; 2) based on some special conditions and prior assumptions (or modelling parameters), some of the ionospheric models are accurate. However, when these conditions and parameters change, the model bias may become large and can make the results unsuitable; 3) some kinds of the models only express the ionospheric distribution in a wide area. It may generate a large bias when they are used in a specific location.
The major contributions of this study are summarised as follows:
(1) Four novel models, which can retrieve electron densities using fewer data sources, are proposed as below:
a) A method is proposed to model the electron density bias (with location dependence) generated by the Abel inversion algorithm. When the calculated bias is removed, the accuracy of the retrieved vertical electron density (VED) can be improved. This method only needs a RO event data to generate a more accurate VED profile.
b) The electron density is firstly modeled with the parameters of altitude and the zenith angle (with regard to the local centre), and then it is estimated by the tomographic ART (Algebraic Reconstruction Technique) algorithms using the GPS data from a RO event and a ground-based receiver.
c) In the ionospheric Shape Function model, the Shape Function is proposed to be altitude and location dependence rather than only altitude dependence (as that in the classic Shape Function model). This Shape Function can be solved by using the GPS data from a RO event and a ground-based receiver.
d) The electron density is firstly modeled with the parameters of altitude, the zenith and azimuth angles (with regard to the local centre, and then it is estimated separately by the tomographic ART (Algebraic Reconstruction Technique), the MART (Multiplicative Algebraic Reconstruction Technique) and the Kalman Filter algorithms using the GPS data from a RO event and a ground-based receiver.
(2) Two quality control methods for TEC modelling are proposed as follows:
a) A method of GPS outlier detection and exclusion in TEC modelling is proposed. Also the impact of outliers on TEC modelling is numerically calculated and analysed.
b) In the estimation of the unknown parameters of the TEC models by a Kalman Filter, the impacts from the modelling parameters are numerically analysed.
(3) A TEC model, which uses a Kalman Filer to smooth pseudo-ranges and to solve for combined biases in the hardware and the model, is proposed. Based on this model, the TEC temporal and spatial variations, which occurred in Australia in 2006, are analysed.